Phase Transformations and Crack Initiation in a High ... · Phase Transformations and Crack...

Phase Transformations and Crack Initiationin a High-Chromium Cast Steel Under Hot

Compression TestsJerome Tchoufang Tchuindjang, Ingrid Neira Torres, Paulo Flores, Anne Marie Habraken, and Jacqueline Lecomte-Beckers

(Submitted November 19, 2014; in revised form February 23, 2015; published online March 12, 2015)

The mechanical behavior of the fully austenitic matrix of high-chromium cast steel (HCCS) alloy isdetermined by external compression stress applied at 300 and 700 �C. The microstructure is roughlycharacterized toward both optical and scanning electron microscopy analyses. Dilatometry is used duringheating from room temperature up to austenitization to study the solid-state phase transformations, pre-cipitation, and dissolution reactions. Two various strengthening phenomena (precipitation hardening andstress-induced bainite transformation) and one softening mechanism (dynamic recovery) are highlightedfrom compression tests. The influence of the temperature and the carbide type on the mechanical behaviorof the HCCS material is also enhanced. Cracks observed on grain boundary primary carbides allowestablishing a rough damage model. The crack initiation within the HCCS alloy is strongly dependent onthe temperature, the externally applied stress, and the matrix strength and composition.

Keywords dilatometry, electron microscopy, internal damage,mechanical characterization, strengthening mechanism,tool steel

1. Introduction

High-chromium cast steels (HCCSs) are tool steels used inapplications where sufficient wear resistance, high hot hard-ness, good oxidation behavior, and high fracture toughness aresought, these properties being also retained at elevatedtemperatures (Ref 1-3). A standard application for HCCS isthe work roll of the roughing stand in the hot strip mill (HSM).In the case of cast work rolls, the initial chemical compositionof the alloy, together with the solidification rate during thecasting route, sets the grain size and the nature of the phasesoccurring in the solidification range, especially primary car-bides often located at grain boundaries while forming a more orless continuous network (Ref 4, 5). These primary carbidesrepresent the main parameter influencing the high hardness andthe wear resistance of the alloy (Ref 4, 6).

The final properties of the material are promoted by thesubsequent heat treatments following the casting route, whichallow the formation of a martensitic matrix containing uni-formly distributed secondary carbides that yield to enhanced

hardness (Ref 4, 7, 8). Though work rolls for the roughing millare massive parts weighing up to 40 tons, the heat treatmentthat follows the casting route must be handled carefully, whileconsidering moderated heating and cooling rates in order toavoid distortions or detrimental residual stresses (Ref 9). Inaddition, soaking times for both hardening treatment andtempering might be long enough to allow, respectively,sufficient homogenization within the as-cast (AC) microstruc-ture and adequate incubation time for carbide precipitation. Thefinal mechanical properties of the work roll strongly depend onboth the casting route and the subsequent thermal treatment(Ref 4-8).

Numerous studies have been undertaken to understand thework roll behavior during hot rolling in order to prevent or tobetter control both surface and internal damage. These worksare related to both experimental and modeling studies, anddifferent damage phenomena are involved, such as bandingfailure, thermal fatigue, fire cracking, wear, etc. (Ref 10). Inaddition, residual stress distribution within the work roll inservice is also taken into account in order to reach accuratesimulations (Ref 11).

Nevertheless, internal damage within the work roll during itsfabrication and prior to service has received little attention sofar.

Actually, it is still difficult to predict the behavior of suchheavy parts, either at the end of the casting route or later duringthe subsequent heat treatment, due to the complex three-levelcoupling between the mechanics, metallurgy, and thermalfields. Specifically, there is a need to know the mechanicalbehavior of every single phase involved in the composition ofthe studied alloy, such as retained austenite, bainite, ormartensite for the HCCS grade. For the existing damagemechanisms, little data are available (Ref 12, 13).

Indeed, specific studies focusing on the hot forging of toolsteels are related either to the higher temperature range between900 and 1100 �C or to the intermediate temperature rangeslightly below 700 �C. The higher temperature range corre-sponds to hot forging processes where dynamic recrystalliza-

Jerome Tchoufang Tchuindjang and JacquelineLecomte-Beckers,Division MMS, Department AME, Universite de Liege, Liege,Belgium; Ingrid Neira Torres, Division MS2F, Department ArGEnCo,Universite de Liege, Liege, Belgium and Department of MaterialsEngineering, Universidad de Concepcion, Concepcion, Chile;Paulo Flores, Department of Mechanical Engineering, Universidadde Concepcion, Concepcion, Chile; and Anne Marie Habraken, Divi-sion MS2F, Department ArGEnCo, Universite de Liege, Liege,Belgium. Contact e-mails: [email protected], [email protected],[email protected], [email protected], and jacqueline. [email protected].

JMEPEG (2015) 24:2025–2041 �ASM InternationalDOI: 10.1007/s11665-015-1464-7 1059-9495/$19.00

Journal of Materials Engineering and Performance Volume 24(5) May 2015—2025

tion, grain refinement, or decreasing primary carbide sizes areexpected (Ref 14, 15). For the intermediate temperature range, athermo-mechanical process also known as ausforming wasdeveloped which aims to plastically deform austenite at aconstant temperature before its transformation (Ref 16). Such aprocess allows more or less dynamic recovery phenomenon, andat the same time, it has been reported to be very effective inrefining the subsequent bainite/martensite structures which maylater improve both strength and toughness (Ref 17, 18).Nevertheless, the two approaches previously quoted do notgive the thermo-mechanical behavior of the austenite phase intool steels. In fact, their main purpose is to achieve forgingconditions that allow maximal strains under reduced stresses inorder to produce a fine-grained structure at room temperature.

In addition, numerous works have been undertaken relatedboth to the theoretical approach and to the simulation of thework hardening phenomenon in metals with a face-centeredcubic (FCC) lattice, such as austenitic stainless steels, austeniticmanganese steels, single crystals, and pure metals having lowstacking fault energy (SFE). However, many of these studiesfocus on twinning, a phenomenon responsible for the subse-quent enhanced ductility obtained on the material (Ref 19-21).Conversely, very few works have been undertaken on materialswith a high SFE, such as HCCS grade material. The HCCS alloystudied in this paper is more like a high SFE material due to thehigh number of alloying elements, the precipitation inside thegrains, and the micro-segregations at the grain boundaries.

The mechanical behavior of ultrahigh carbon steels atelevated temperatures is also investigated for various high strainrates to meet the requirements for developing dip die forging andother thermo-mechanical processing operations involving highdeformations (Ref 22, 23). However, these studies deal with highSFE austenitic alloys. They mostly focus on the optimization ofmaterial formability through the construction of the deformationmechanism maps and the determination of the maximumductility. Nevertheless, the main objectives of the above-mentioned works are quite far removed from the enhancementof the strengthening behavior of complex austenite phase withhigher SFE, such as the HCCS involved in this work, in theintermediate and especially in the lower temperature ranges.

The current study forms a basis for the subsequentsimulation of the thermo-mechanical behavior of the shellmaterial of cast work rolls which contain different types ofcarbides (Ref 9).

The strengthening behavior of austenite in the studiedHCCS alloy combines different hardening and softeningphenomena. The current analysis enhances their specificcontribution to the overall material behavior. This studychooses both an intermediate temperature and a low tem-perature for the compression tests performed together with twoheating rates that allow obtaining more or less undissolvedtransition carbides within the matrix.

2. Material and Experimental Methods

2.1 Raw Material and Sampling

The studied alloy is an HCCS grade originating from theshell material of a work roll obtained from a vertical spincasting process and dedicated to the roughing stands of anHSM. The work roll has an outer diameter of 1200 mm and ashell depth of 80 mm. The average chemical composition of the

HCCS alloy as obtained from optical emission spectroscopy isgiven in Table 1.

Rough cylindrical samples with their axes parallel to that ofthe work roll are cut out within the first 20 mm over the entire80-mm depth of the shell material. The samples are locatedclose to the end of the barrel. The sizes of the cylindricalsamples range between 5 and 10 mm for the diameter, andbetween 5 and 25 mm for the length, depending on the final testto be performed (Fig. 1).

The cylindrical specimens are obtained from an electro-discharge machining process. A preliminary metallographicpreparation consisting of a hot mounting with epoxy resinfollowed by grinding and polishing down to 1 lm is performedprior to final super polishing down to 0.06 lm with colloidalsilica, which is an active oxide polishing suspension (OPS).Elsewhere, classical nital etching is undertaken on stressed andpolished samples in order to better reveal the phases within thematrix and to enhance transformation of parent austenite intohardened phases (bainite, martensite).

2.2 Experimental Methods

Heat treatment is undertaken on cylinders coming fromHCCS blocks in the AC conditions, to allow both re-homogenization and hardening in the studied alloy. The so-called HT (heat-treated) samples are generated from this heattreatment as illustrated in the flow chart process given in Fig. 1.HT samples are the starting point for further experimentalanalyses, such as dilatometry or compression tests.

The compression tests are carried out with a SCHENCKHydropuls 400 kN machine using a quad elliptical heater49 2000 W, with a constant strain rate of 39 10�3 s�1. Twoheating rates are used on the HT samples when reheating themup to 950 �C: a high heating rate of 2 �C s�1 (A cycle) and alow heating rate of 1 �C s�1 (B cycle). Once the austenitizingtemperature is reached, the cooling stage is obtained inside thefurnace down to the temperature of the compression test, whichis either set at 700 or at 300 �C. For a given heating rate A or B,each compression test is performed twice. In addition, in orderto analyze the pure effect of the thermal cycle on themicrostructure in the furnace used for compression tests, thethermal cycle B is run on a single additional sample withoutapplying any compression stress.

Microstructure characterization is carried out at differentstages of the work using various techniques such as opticalmicroscopy (OM), scanning electron microscopy (SEM), andVickers hardness measurements.

OM observations are made on an OLYMPUS BX60 micro-scope coupled with an OLYMPUSUC30CCDdigital camera andequipped with a motorized stage together with the OLYMPUSSTREAM Motion� software. OM analyses are performed onlongitudinal sections of cylindrical samples originated from AC,HT, and compression test (CT) conditions (Fig. 1).

SEM analyses are performed on a PHILLIPS XL 30 FEG-ESEM equipped with a 129 eV BRUKER energy dispersive x-ray spectrometer (EDS) with a silicon drift detector (SDD), thesystem being connected to ESPRIT� software for pre- and post-treatment of the data. The SEM allows identification of thematrix phases and carbides greater in size than one micron.Such a technique is not appropriate for thin precipitates as smallas some nanometers.

The transmission electron microscopy (TEM) technique,often used for both dislocations and nanoscale precipitate

2026—Volume 24(5) May 2015 Journal of Materials Engineering and Performance

studies, was discarded due to difficulties in preparing thin foilsamples. With the available electro polishing device, carbideswithin the HCCS alloy remain undissolved during the prepa-ration. Instead, dilatometry tests are carried out on HT samples,on a NETZSCH DIL 402C apparatus. Dilatometric tests consistof a reheating stage up to 960 �C prior to cooling down to roomtemperature. Two constant heating rates of 2 and 20 �C min�1

are used, while a single cooling rate of 20 �C min�1 is run. Thedilatometry technique is used to detect tertiary carbides withinthe HT HCCS material, and to measure the critical temperaturesfor their precipitation and their subsequent dissolution. Inaddition, such a test helps to determine the coefficient of thermalexpansion (CTE) of the phases existing in the studied sample.

Hardness measurements are done on a universal EMCOMC10 010 machine with an electronic cell force and closedloop regulation. Both micro-(HV1 for 1 kg load) and macro-(HV30 for 30 kg load) Vickers hardness are obtained in orderto, respectively, assess the matrix hardness and the bulkhardness. For each sample, the average hardness value isobtained from 3 or 5 indentation points, respectively, for themicro-hardness (inside single grains) and for the macro-hardness (bulk hardness for about twenty grains includingcarbides located at grain boundaries).

Grain size measurements and primary carbide volumefractions are obtained using image processing software on

compressed specimens. Grain size is obtained following theplanimetric method (ASTM E112 chart). The carbide volumefraction is assessed while considering all the grain boundarycarbides as a whole. The results of carbide quantification mayrepresent the upper limit for their actual volume fraction withinthe studied samples. The main parameters for the setting of thequantitative metallography (grain size measurements andcarbide quantification) are given in Table 2.

Sample designations with the related applied thermal cycleand heating rate are summarized in Table 3. Hereinafter, thefollowing designations AC, HT, CT300, and CT700, will beused, respectively, for the as-cast sample, the heat-treatedsample, the sample re-austenitized prior to the compression testat 300 �C, and the sample re-austenitized prior to the com-pression test at 700 �C. For the compressed samples, anadditional letter, A or B that refers to the heating rate, can beused. The sample re-austenitized and cooled down to roomtemperature without applying any stress will be called thestress-free sample or SF300.

3. Results

In this work, the morphology, the distribution, and thechemical composition of the so-called primary and secondary

Table 1 Average chemical composition (weight %) of the HCCS alloy

Alloying elements

C Si Mn Ni Cr Mo V Fe

Min 1.0 0.2 0.6 0.5 10.0 3.0 0.1 BaseMax 1.5 1.0 1.0 1.5 14.0 6.0 0.5

Fig. 1 Flow chart for the sampling and the experimental tests on HCCS alloy


carbides are mentioned in section 3.1. In addition, the criticaltemperatures for both the primary and the secondary carbidesare measured according to the specific nomenclature, deter-mined in a previous study (Ref 24), reproduced hereafter:

• Primary carbides are eutectic types composed of Cr-richM7C3 or Mo-rich M2C that precipitate from the liquid atgrain boundaries during the solidification process. Theyexhibit both lamellar and rod-like morphologies. Theirmain sizes range between 5 and 50 lm and they form at asingle temperature for each, which is above 1100 �C.

• Secondary carbides precipitate around 1050 �C from the su-persaturated austenite, in continuous cooling conditionsduring the casting route and just after the solidification pro-cess. These carbides are composed of Cr-rich M23C6 wide-spread inside grains, which define, at the same time aroundtheir cluster, a band strip also known as the ‘‘precipitate-freezone’’ (PFZ) that is close to the grain boundaries. They aremainly globular shaped, with sizes between 1 and 3 lm.

The mechanical results, microstructure characterization, andhardness and quantitative metallography are, respectively,given in sections 3.2, 3.3, and 3.4. Discussions involvingvarious strengthening and softening mechanisms and theinternal damage phenomena occurring within the compressedHCCS alloy are addressed in section 4, together with thedevelopment of a rough damage model. In particular, twoextended studies based on the literature and the current resultsare proposed. Sections 4.2 and 4.3 provide, respectively, for thebainite transformation and for the carbides, the strengtheningmechanism that occurs in the HCCS alloy.

3.1 Solid-State Transformations and Precipitation/Dissolution Reactions from the Dilatometry Study

Microstructures in the AC conditions, and after the re-homogenization and hardening heat treatment, are illustrated inFig. 2.

The microstructures in both the AC (Fig. 2a) and HT(Fig. 2b) samples are quite similar, with a quasi-continuous

network of carbides at grain boundaries and a mixed matrixcomposed of both fresh martensite (a¢) and retained austenite(c). In addition, troostite is only present in the AC samples(Fig. 2a). The heat treatment undertaken on the AC samplesallows full dissolution of the troostite phase into austenite priorto the subsequent martensitic transformation during air cooling.The final microstructure obtained on the HT samples exhibitsenhanced bulk hardness as it is mainly composed of amartensitic matrix with a small amount of retained austenite(Table 4). Primary and secondary carbides are assumed toremain similar in the AC conditions and after re-homogenizingtreatment because the austenitization temperature chosen isbelow the critical temperature at which the dissolution ofsecondary carbides begins.

The chemical distribution of elements within both thecarbides and the matrix as obtained from the SEM-EDSanalysis of the HT samples is illustrated in Fig. 3.

It appears that both primary Cr-rich M7C3 and Mo-richM2C carbides are mainly large eutectic carbides located atgrain boundaries in a quasi-continuous network. Thesecarbides exhibit constant chemical composition for eachtype. There are intragranular secondary Cr-rich M23C6

carbides that form clusters inside grains, with a chemicalcomposition that changes depending on the carbide location.The Mo content in secondary carbides seems to be higher forprecipitates located close to the grain boundaries than those inthe center of the grain. In addition, the M23C6 carbide clustersleave around and close to the grain boundaries the so-calledPFZs that describe annular strips without visible carbidesinside.

Dilatometric curves obtained on the HT samples during thereheating stage up to 960 �C under two constant heating ratesof 2 and 20 �C min�1 are given in Fig. 4, with the illustrationof the CTE related to the phases that are present within thematrix.

Both HT samples exhibit a small expansion well below theAc1 point, which corresponds to fresh martensite (a¢) decom-position that also involves the formation of ferrite (a) andcementite (h). This phase transformation occurs earlier under alow heating rate compared to the high heating rate. In addition,

Table 2 Samples and setting used for grain size and primary carbide volume fraction assessment toward opticalmicroscopy

Sampledesignation

Overall sectionsize (mm)

Orientation ofthe section

Frame resolution(pixels)

Area/frame(mm2)

Total numberof frames

Ratio of the sampledregion over the total area (%)

CT300-A Height 5Width 9

L(// roll axis)

2592 x 1944 5.8 3 12.9CT300-BCT700

Table 3 Summary of the thermal cycles applied during reheating up to 950 �C, and temperature of the subsequentcompression tests performed during the subsequent cooling stage inside the furnace

Temperature forthe compression tests Sample designation

Number ofsamples

Heating rate achieved up to950 �C, prior to compression test

300 �C CT300-A 2 2 �C/sCT300-B 2 1 �C/s

Stress-free SF300-B 1 1 �C/s700 �C CT700-A 2 2 �C/s

CT700-B 2 1 �C/s


the thermal expansion of the reheated HT samples seems to behigher with the low heating rate. Moreover, Ac1 and Ac3corresponding, respectively, to the start point and to the endpoint of the austenite transformation are shifted to highertemperatures on increasing the heating rate.

Solid-state transformations with their related critical points,which are clearly identified on the dilatometric curves, maycorrespond to the following sequence: the a¢-fresh martensitedecomposition, the c0-austenite transformation, and the disso-lution of h-cementite into c0 to yield c1-austenite well aboveAc3. The later reaction is only observed on the HT sample withthe low heating rate. The subscript (i = 0 or 1) used foraustenite (ci) is intended to illustrate the difference within thechemical composition of the related phase. In fact, c1 thatresults from the dissolution of h into c0 is more saturated in Cthan the parent c0 phase that comes from the transformation ofa.

The increase in the heating rate seems to delay both thea¢-martensite decomposition and the c0-austenitization range(Ac1 and Ac3), to higher temperatures.

Dilatometric curves exhibit three consecutive steady stagesI, II, and III, which are represented by three constant slopes,each stage having a constant CTE that may be related to one ormore ‘‘stable’’ phases (Fig. 4). The first stage is related to a¢-martensite phase (a small amount of retained austenite can bepresent), the second to both a-ferrite and h-cementite, and thethird to c0-austenite. The first two stages are below the Ac1

point, while the third is above the Ac3 point. Any change in theslope of the curve corresponds either to a phase transformationor a precipitation/dissolution reaction. The corresponding CTEof a¢ ranges between the lowest one subsequently obtained ona + h, and the highest one achieved on c1, at the end of theheating mode (Fig. 4). The decrease in the CTE of the newlyformed a-ferrite compared to that of a¢-martensite is expectedas the tetragonal crystal lattice of the martensite is moresaturated in interstitial atoms than the body-centered cubiclattice of the ferrite.

The decomposition of the a¢-martensite into a-ferrite and thenewly precipitated h-cementite yields a small expansion in thedilatometric curve, while the allotropic transformation of a-ferrite into c0-austenite occurs under significant contraction(Fig. 4).

Moreover, the HT sample reheated under the low heatingrate seems to exhibit a CTE slightly higher than that of thesample with a high heating rate, just above the Ac3 point(Fig. 4). This phenomenon is well enhanced when coolingcurves are considered for both samples due to the differencebetween the initial thermal expansion of the c-austenite phaseduring the heating mode, and the corresponding contractioncoefficient of the same phase during the subsequent coolingmode. Such a difference only exists for the HT sample with thelow heating rate (Fig. 4), and it may correspond to h-cementitedissolution into c-austenite. This result is in good agreementwith Chae et al. (Ref 25) who found that cementite dissolution

Fig. 2 Microstructure (optical micrographs) of the HCCS alloy in the as-cast conditions (AC samples) and after heat treatment (HT samples)

Table 4 Average Vickers hardness of the HCCS alloy after various thermo-mechanical treatments

Sampledesignation

Macro-hardnessmeasurements (HV30)

in bulk material

Micro-hardness measurements(HV1) with a single indentation

inside each grain

Grains with bainite sheaf structure Grains free of bainite sheaf structure

AC 582± 7 … …HT 805± 9 … …CT300-A 622± 4 599± 22 634± 0CT300-B 664± 8 653± 4 685± 8CT700-ACT700-B

652± 6 635± 2 649± 3

SF300-B 613± 10 603± 12 640± 16


in the austenite matrix under continuous heating conditionsyields non-linear behavior on the dilatometry curve.

Therefore, it can be concluded that the matrix of HT samplesat 960 �C is composed of c0 + undissolved h-cementite whenthe reheating rate is as high as 20 �C min�1. The same samplesreheated at 2 �C min�1 may exhibit a matrix composed of asinge c1 phase, where h-cementite precipitates are almost fullydissolved.

3.2 Flow Stress Curves and Work Hardening Rates of theHCCS Alloy

The thermal histories measured during the time spent by thesamples in the furnace for the CTs are illustrated in Fig. 5(a)below. The related flow stress curves obtained from the CTs(logarithmic strain-true stress) are given in Fig. 5(b).

Flow stress curves show a hardening stage that increases withdecreasing compression temperatures. All the samples subjectedto the low heating rate of 1 �C s�1 prior to the CT, performedeither at 700 or at 300 �C, exhibit similar flow stress curves.Therefore, the average between both curves is calculated and isconsidered hereinafter. In addition, flow stress curves obtainedfrom the CT700-A samples are almost identical to the average ofthe CT700-B samples. This suggests similar hardening behaviorfor the batch of samples tested at 700 �C, regardless of the priorreheating rate up to the austenitizing temperature. The foursamples tested at 700 �C will be referred to as CT700hereinafter, irrespective of the heating rate used prior to the CTs.

Conversely, different mechanical behavior is observedbetween the two samples reheated at 2 �C s�1 (CT300-A)

and the two other samples reheated at 1 �C s�1 (CT300-B)prior to the CTs performed at 300 �C (Fig. 5b). The CT300-Asamples exhibit the higher hardening effect, while maximumelongation has been achieved on the CT300-B samples.

The work hardening rate curves of all the tested samples areillustrated in Fig. 6. The work hardening rate curves present adecreasing trend below 0.02% logarithmic strain, beforereaching the same asymptote above this limit. The decreasingtrend in the work hardening rate is higher for the CT300-Asamples, probably because the maximum strength reached ismany times higher than that of the other samples.

3.3 Microstructures After the CTs

The microstructures obtained after the CTs performed at300 �C are illustrated in Fig. 7 and 8, respectively, for the high(CT300-A) and for the low (CT300-B) heating rates.

The CT300-A samples contain bainite with preferentialoriented sheaves, the rest of the matrix being composed of freshmartensite and possibly retained austenite (Fig. 7a). The bainitesheaf structure probably resulted from the compression stressapplied to the sample. The crack path observed in the CT300-Asamples is composed of single oriented cracks within the grainboundary carbides, these cracks being parallel to the externallyapplied stress (Fig. 7b). In addition, evidence of the secondarycarbide�s widespread inside grains is enhanced in the opticalmicrograph after OPS preparation (Fig. 7c).

The CT300-B samples exhibit bainite with oriented sheavesin their matrix, and either martensite or some retained austenitein addition (Fig. 8a). The crack path in the CT300-B clearly

Fig. 3 EDS-map of HT samples enhancing carbide distribution and composition


looks more complex and more extended than the previous oneobserved in CT300-A, due to the branching that occurs withinthe cracks (Fig. 8b). But the primary cracks remain parallel tothe stress direction. Similar to CT300-A, evidence of numerousintragranular secondary carbides is enhanced in the opticalmicrograph after OPS preparation (Fig. 8c). In addition, thecorrosion pits observed on the CT300-B are probably due to theinfiltration of the nital etching into the large cracks that existwithin the primary carbides (Fig. 8a). PFZs still exist on theCT300 samples (Fig. 7c and 8c).

For the CT700 samples (Fig. 9), very few large orientedbainitic sheaves are observed inside the grains, the rest of thematrix being composed of martensite and retained austenite

(Fig. 9a). This later phase is probably present in an amounthigher than that of the previous CT300 samples. Such anassumption can be supported by two observations. First, theapparent volume fraction of the bainitic sheaves in the CT700samples seems to be lower than that observed within the CT300samples. Second, the low sensitivity of the etched sample to thenital reagent suggests that the austenite phase which remainedun-etched is present in a significant amount.

While all the compressed samples seem to exhibit prefer-ential oriented bainitic sheaves in various amounts and sizes,the stress-free SF300 sample contains large quantities ofsheave-like bainite structures with haphazard oriented direc-tions inside the grains (Fig. 10a). In addition, the grain

Fig. 4 Enhancement of the effect of heating rate on the phase transformations of HT samples during dilatometry tests

Fig. 5 (a) Thermal histories of the compression samples (b) Flow stress curves obtained after compression test performed during cooling stageeither at 700 or at 300 �C


boundary carbides do not exhibit cracks as no external stresswas applied. Widespread secondary carbides are present insidethe grains with PFZs around them (Fig. 10b), such anobservation being similar to the previous ones made on thecompressed samples.

3.4 Hardness Measurements, Grain Size, and CarbideContent

Hardness values are given in Table 4 with bulk hardnesscorresponding to macro-hardness, and micro-hardness corre-sponding to local hardness inside the grains. Hardness mea-surements are located in two different zones or grains, the firstone containing bainite sheaf structure and the second one beingfree of bainite. This latter zone probably corresponds to themartensite phase with very low retained austenite. The averagemacro-hardness ranges between the two extreme values foundfor micro-hardness inside the grains, except for the CT700samples where it is close to the higher micro-hardness valuesobtained.

The maximum macro-hardness is achieved on the HTsamples, which contain a higher amount of martensite with areduced quantity of retained austenite. In addition, no bainiticphase is found in these samples.

The lower bulk hardness value is obtained on the ACsamples, probably due to the presence of the softer troostitephase (Fig. 2a).

Among the micro-hardness values, the lowest ones thatcorrespond to 599 HV1 and 634 HV1 are obtained for theCT300-A samples, respectively, within grains containingbainitic sheaves and in grains without bainite phase. TheCT700 samples exhibit hardness values that are insertedbetween those of the CT300-A and CT300-B for both macro-and micro-hardness. For the SF300-B sample, the minimum ofthe micro-hardness (603 HV1) is observed inside the grainscontaining bainite sheaf structure, while the maximum of 640HV1 is obtained inside the bainite-free grains.

The maximum for the standard deviation has been measuredwithin the CT300-A samples, probably due to very largediscrepancies within the progress of the bainite-orientedtransformation inside the grains.

The hardness measurements for the compressed samplesconfirm the strong differences within the HCCS behavior asobserved in both the AC and HT samples. Section 4.5 will linkthese results to the microstructures generated during theheating, cooling, and, finally, the compressed states.

The results of the grain size and carbide content assessmentsare given in Table 5. Both the grain size and the volumefraction of the primary carbides are similar for all thecompressed samples after the CTs.

4. Discussion

4.1 Work Hardening Rate in Compressed Samples

The work hardening rate curves of all the tested samplesrepresent Stage III and Stage IV (Fig. 6) as described byKalidindi for FCC polycrystals with high SFE, for whichplastic deformation proceeds by slip alone, contrary to low SFEFCC metals (plastic deformation by both slip and twinning)(Ref 20). In addition, the first two stages (Stage I and Stage II)predicted by Kocks and Mecking (Ref 21), which are welldefined for single crystal materials with low SFE, are reportedto be hardly observed for FCC polycrystals with high SFE,which are similar to the HCCS alloy studied in this work.

In polycrystalline metals deforming by slip alone, the strainhardening response primarily comprises a dynamic recoveryregime (Stage III) in which the strain hardening rate con-tinuously decreases with increasing flow stress or imposedstrain (Ref 20). At high strain rates, Stage III of such alloys mayturn into Stage IV which corresponds to the region for lowhardening rates where an asymptote for high strain is set(Ref 20, 21).

Fig. 6 Work hardening rate curves for HCCS material at 300 and 700 �C


Fig. 7 Microstructures obtained after compression tests onCT300-A samples Fig. 8 Microstructures obtained after compression tests on CT300-

B samples


For the CT300-B samples which have a similar but moreexpanded Stage III than that of the CT700 samples, both thelow heating rate up to the austenitizing temperature and therelatively longer holding time at this temperature promoteh-cementite dissolution into the fully c1 austenitic matrix. Thecomposition of the parent c0 austenite phase is assumed to bemodified at the same time, as all the elements previouslycontained in h may enter into the composition of c1. Newly

dissolved elements such as transition elements (Cr, Mo, and V),or interstitial elements such as C, respectively, act as substi-tution or insertion elements inside the FCC austenite lattice,which later promote the solution strengthening effect.

In a ‘‘sub-saturated’’ c0-austenite containing undissolvedtertiary carbides (h) such as that of the CT300-A samples, thework hardening behavior should be more influenced by theOrowan mechanism that involves dislocation-precipitate inter-

Fig. 9 Microstructures obtained after compression tests on CT700 samples

Fig. 10 Microstructures obtained on the stress-free sample (SF300)

Table 5 Grain size and carbide volume fractions obtained on various samples in the longitudinal direction, in differentconditions

Sample designation Grain size (G) according to ASTM E 112 Carbide volume fractions (%)M7C3 + M2C

CT700 7.3± 0.1 18.6± 0.4CT300-A 7.5± 0.1 17.4± 1.5CT300-B 6.6± 0.5 19.3± 0.7


actions in combination with either a phase transformationstrengthening mechanism or a softening dynamic recoveryphenomenon (Ref 21).

Finally, HCCS behavior at 700 and 300 �C fits the abovedefinitions of Stages III and IV very well, except for theCT300-A samples, which exhibit Stage III well above that ofthe other samples. This difference can be explained through themicrostructure itself (especially fine tertiary carbides, seesection 4.3.1), as it is established that the preliminary reheatingand cooling sequence prior to the CT can strongly influence thecomposition of the matrix.

4.2 Kinetics of Stress-Induced Bainite Transformation(SIBT)

4.2.1 Effect of Temperature and Stress on the Initiationof the SIBT. It is known that critical points for phasetransformations strongly depend on the chemical compositionof the parent austenite that may change the critical phasetransformation temperatures. A former study (Ref 26) of asimilar grade suggests that the bainitic nose in the continuouscooling transformation (CCT) diagram is far enough to preventbainitic transformation for the applied cooling rate. However,this assumption is inaccurate according to the currentmicrostructure observations (see section 3.3). In addition, theuniaxial applied stress can increase the start of the bainiticreaction while hindering the transformation of non-favorableoriented grains at the same time (Ref 27, 28).

Optical micrographs of samples that have undergone CTseither at 300 �C (Fig. 7a and 8a) or at 700 �C (Fig. 9a) showtypical bainitic sheaves oriented following preferential direc-

tions, contrary to the stress-free samples (Fig. 10a), whichexhibit haphazardly distributed sheaves of bainite. At the sametime, the microstructure of the samples tested at 700 �C isdifferent from that of the samples tested at 300 �C. In the formersample, bainitic sheaves are thicker but in relatively low quantitywith a more limited number of orientations. These results are ingood agreement with Hase et al. (Ref 27) who found that stress-assisted transformation results in large blocks of bainite inidentical orientation, the sheaves growing under the influence ofstress being thicker than those occurring without stress. Inaddition, the fact that bainitic sheaves look thicker in the CT700samples than in the CT300 specimens also suggests a ‘‘coars-ening’’ phenomenon that occurs on the sheaf-like microstructuredue to the higher transformation temperature (Ref 27).

4.2.2 Effect of Temperature and Stress on the Growth ofthe SIBT (Amount of Retained Austenite). The retainedaustenite seems to be present in a higher amount in the CT700samples than in the CT300. The fact that the apparent amountof bainitic sheaves seems to decrease with the increasingtemperature of the CTs seems to be consistent with previousworks (Ref 29-32), where it is shown that prior strain or priorstress of the parent austenite greatly influences the subsequentphase transformations during the cooling stage, especially thebainitic transformation.

Yang et al. (Ref 29) and Liu et al. (Ref 30) reinforce the factthat the prior deformation of austenite allows its mechanicalstabilization in such a way that the subsequent bainitictransformation is significantly delayed, respectively, in isother-mal and in continuous cooling conditions.

Jin et al. (Ref 31) found that the isothermal decompositionof deformed austenite into bainite is significantly promoted as

Fig. 11 Sketch of the strengthening and softening mechanisms and the internal damage process (crack initiation on grain boundary primarycarbides) occurring on the HCCS alloy at different temperatures


the incubation period is shortened with increasing strain rate. Inthe same study, it is noted that the start of isothermal bainiteformation increases when the parent austenite phase is de-formed. Most recently, Gong et al. (Ref 32) have found that theplastic deformation of the austenite phase accelerates thebainite transformation at a low temperature (300 �C, for amedium alloy steel), whereas the effect on the same transfor-mation is reduced at a higher temperature (600 �C). Theyestablished that the difference between the two temperatures isascribed to the dislocation structure introduced in the parentaustenite phase, which remains on the active slip planesespecially at the lower temperature. These dislocations laterassist the bainite transformation under strong variant selection.

Nevertheless, the above findings may not be consideredcontradictory if additional information is taken into account.Bhadeshia et al. (Ref 27, 28, 33-35) found that the compressiveuniaxial stress increases the bainite start temperature, acceler-ates the transformation rate, and causes microstructural align-ment presented by typical sheaves in selected crystallographicvariants of bainite preferentially formed on planes of maximumshear stress. In the meantime, the non-favorable directions arehindered by the same external stress because the shapedeformation of a bainite plate does not comply with theexternal stress; therefore, they should remain austenitic.

Retained austenite that is weakly etched by nital reagentseems to be still present in the CT700 samples (Fig. 9a), in anamount which is lower than that of both the CT300-A (Fig. 7a)and CT300-B samples (Fig. 8a). In addition, the typicallyoriented bainitic sheaves observed in the final microstructure ofthe CT700 samples are larger than those found in the samplescompressed at 300 �C, thus confirming their origin from thestressed material. Note that stress-free deformed austenitegrains do not allow variant selections for bainite transformationand yield to haphazardly distributed sheaves of bainite (Ref 36)as observed in the SF300-B sample (Fig. 10a).

4.2.3 Conclusions on SIBT. It seems that the stressapplied at both 300 and 700 �C allows bainitic transformationunder variant selection. The retained and deformed austenite at700 �C exhibits a mechanical stabilization that avoids furthertransformation into bainite, explaining why the volume fractionof bainite formed at 700 �C due to external stress is lower thanthe amount of bainite formed at 300 �C.

In addition, higher strains are observed on the samplescompressed at 300 �C, especially when the specimens werepreviously submitted to a low reheating rate up to theaustenitization temperature. Such large strains lead to a highamount of bainite that is issued both from the conventionalphase transformation, which is linked to thermal history, andfrom the SIBT.

4.3 Carbide Influence on Hardening and DamageMechanisms at 300 �C

4.3.1 Transition or Tertiary Carbides. It is important tospecify why tertiary carbides are called either transitioncarbides or h-cementite (M3C) in the present work.

First, if isothermal conditions such as those achieved duringthe tempering heat treatment of high alloyed steels areconsidered, then we are dealing with a moderate heating rateup to a temperature well below Ac1 with holding times fromtens of hours. The long exposure times are needed both to allowan adequate incubation time for the so-called ‘‘secondary’’carbide precipitation into the tempered martensitic matrix, and

to promote further destabilization of retained austenite duringthe subsequent cooling (Ref 2, 6-8, 37-39). For long agingtimes, ‘‘secondary carbides’’ may also undergo transformationinto species that are more stable depending on both thetemperature and the alloying elements. According to Durand-Charre (Ref 37), the first carbide to precipitate from themartensitic matrix in the range 500-600 C during the temper-ing of alloy steels is h-cementite (M3C), prior to a partialdissolution that allows carbon released and further combinationwith other elements whose carbides are more stable. The sameresult was obtained by Shtansky et al. (Ref 38), whose studyfocuses on the decomposition of fresh martensite just belowAc1 on an Fe-17Cr-0.5C alloy. In almost all the previousstudies, the initial precipitates are often strongly coherent withthe matrix, with well-established orientation relationships andwith their nuclei being only a few nanometers in size. Thecoherency with the matrix may be gradually lost withprolonged high temperature exposure due to a possiblecoarsening phenomenon, especially for carbides with complexcrystal structures and low heat for formation (M7C3, M6C, andM23C6). Therefore, h-cementite (M3C) precipitating duringtempering under isothermal conditions can be considered asnuclei for more stable carbides which appear later within themartensite matrix, at the expense of the primary h-cementite.

As a second step, if continuous heating conditions up to thecompletion of the austenite transformation are considered, thenphase transformations within the matrix of the HCCS alloy willrange between 570 and 860 �C (see the dilatometric curve inFig. 4). Just below Ac1 and prior to austenite transformation,the fresh martensitic matrix will decompose into ferrite and h-cementite (M3C) (Ref 37, 38). Between the points Ac1 and Ac3,the ferrite matrix will be transformed into austenite under anallotropic transformation, while h-cementite will remain mostlyundissolved or untransformed, due to the continuous heatingconditions (Ref 25, 38, 39).

As the continuous heating conditions of the HCCS alloy arefar away from the equilibrium, the complete transformation ofmetastable cementite into more stable Cr-rich carbides mayoccur neither below Ac1 nor later during the austenitetransformation. Therefore, it has been assumed that h-cementiteremains present and untransformed within the matrix eitherafter its nucleation in the fresh martensite, or later whenprecipitating from the decomposed martensite, below Ac1.

At the end of austenite transformation, the HCCS alloy isassumed to be composed of a fully austenite matrix containingultrafine carbides mainly made of undissolved h-cementite(M3C), and possible ‘‘stable’’ carbides coming from thedecomposition of partially dissolved h-cementite. Above Ac3,h-cementite (M3C) and other species will undergo dissolutionwithin austenite in a temperature range whose width variesdepending both on the chemical composition of the carbides,and on the heating rate. In fact, Cr is known to enhancecementite stability at high temperature, because retainedcarbide dissolution in austenite is not controlled by highdiffusion rates of interstitial carbon, but by low diffusion ratesof substitutional elements at the interfaces (Ref 25, 37-39). Inaddition, high heating rates may promote the so-calledcementite remnants, which consist of undissolved particleswithin the austenite matrix up to 1000 �C (Ref 40).

From the above discussion, tertiary h-cementite is referred toas transition carbide because it can be transformed into morestable species depending on the reheating route. In addition,h-cementite (M3C) that appears from fresh martensite decom-


position prior to austenite transformation is clearly differentfrom the one which precipitated earlier within the temperedmartensitic matrix, at a lower temperature. The cementite thatprecipitated later can remain undissolved within the austenitematrix above point Ac3, due to both the high heating rates andthe enhanced stability by substitutional elements partitionedinto cementite.

As a consequence, transition carbides remained ultrafine andcoherent with the c0-austenite matrix under a high heating rateup to 950 �C (A samples). Conversely, at least the coarseningof these tertiary carbides with a loss of their coherency with theaustenite matrix is achieved under low heating rate conditions(B samples). Otherwise, partial to complete dissolution of thesecarbides might occur to yield supersaturated c1-austenite. Suchan assumption of the increase of the dissolution temperature ofthe transition carbides with the heating rate was well estab-lished in a previous dilatometric study (Ref 41).

In conclusion, the reason for the higher strengthening effectobserved at 300 �C for the CT300-A samples when comparedto the CT300-B samples may be related to the presence ofundissolved ultrafine and coherent transition carbides in theformer samples that should heavily impede dislocation glideunder the Orowan mechanism (Ref 21, 24). These undissolvedcarbides are assumed to be mainly composed of h-cementite(M3C) that precipitated first. They are also assumed to be stabledue to both the presence of Cr and to the continuous heatingconditions that delay transformations into other stable carbides.Similar transition carbides might have been more or lessdissolved in the CT300-B samples due to their low reheatingrate. Finally, the difference in total elongation between theCT300-A and CT300-B samples is linked to the fact that thegrains of the former, which are full of undissolved tertiarycarbides, are less deformable than those of the latter.

4.3.2 Secondary Carbides. Secondary carbides presentin the HCCS alloy consist of M23C6 carbides that arewidespread inside grains while leaving the PFZs away andclose to the grain boundaries (Fig. 3, 7c, 8c, 9b, and 10b).These carbides are enriched in Cr but their Mo contentincreases close to the grain boundaries. The secondary carbidesare assumed to enhance the overall hardness of the matrix (Ref2). In addition, secondary carbides are also assumed to bepresent in a similar amount in all the compressed samples asthey remain undissolved after the reheating stage up to 950 �Cprior to compression, regardless of the reheating rate previouslyused. But contrary to tertiary carbides, the secondary carbidesmay have little influence on the strengthening effect due to thevery few coherency relationships with the newly formedaustenite.

4.3.3 Primary Carbides. Primary carbides are made ofCr-rich M7C3 carbides and Mo-rich M2C carbides, which arelocated at grain boundaries. But only the M7C3 carbides, whichrepresent the majority of the grain boundary carbides, containmore or less cracks that are related to the total elongationachieved under the external compression stress applied(Fig. 7b, 8b, 9b). In addition, the maximum elongation undercompression at 300 �C seems to be related to the progress ofthe internal damage, as the CT300-A samples exhibit a totalelongation lower than that achieved on the CT300-B samples.

Focusing on the crack path, it is made of primary carbidescontaining cracks parallel to the external compression stress,which correspond to the beginning of the internal damageprocess (CT300-A samples, Fig. 7b). With increasing strain,the crack path changes to a more complex oriented one

(CT300-B samples, Fig. 8b). Therefore, it could be assumedthat the primary cracks parallel to the external stress correspondto the initiation of the internal failure mechanism, which is ingood agreement with Dunckelmeyer et al. (Ref 42). They foundthat, for similar carbides, primary cracking is always parallel tothe loading direction. The crack occurs when the normal stressof the carbide induced by the plastic deformation of thesurrounding matrix is exceeded. This can be achieved eitherfrom dislocation pileups on the grain boundary carbides(Ref 43, 44), or due to micro-stresses resulting from differencesbetween the thermal and the elastic properties of the matrix onthe one hand, and of the eutectic carbides on the other (Ref 45).

In addition, Margolin et al. (Ref 46) in an earlier study thatfocuses on brittle fracture mechanisms set criteria for cracknucleation in the form of micro-discontinuities within thematerial under external stress. In most cases, the crackpreferentially nucleates on a grain boundary carbide becausethe strength of such a phase is lower than the strength of thecarbide/matrix interface. The micro-crack occurs in a carbidewhen the local stress at the head of the dislocation pileupexceeds the strength of the carbide, while remaining lower thanthe more resistant carbide/matrix bound at the same time. Thesame authors also found that the crack nuclei change fromcarbide micro-cracking to delamination at the carbide/matrixinterface, only if higher stresses are reached for the sameloading. Then a pore is formed at the carbide/matrix bound dueto the existence of anti-parallel dislocation pileups within theinterface, due to the higher local stresses. Moreover, Kroon andFaleskog (Ref 44) in a more recent work on ferritic steelsestablished that the cleavage fracture is often initiated on brittlecarbide as a result of a fiber loading mechanism in which thestress levels in the carbides are raised while the surroundingferrite undergoes plastic deformation.

The other cracks especially observed on the CT300-Bsamples, which are oriented in directions different from that ofthe external stress, are probably ascribed to the progress of theinternal damage process with the micro-plastic strain evolution,which allows the occurrence of other critical shear directionswithin the carbides.

In conclusion, the primary grain boundary M7C3 carbidesserve as preferential crack initiation sites under externalcompression stress. The cracking of the carbides occurs dueto their brittleness as they do not allow plastic deformationcontrary to the surrounding matrix.

4.3.4 Conclusions on the Carbide Influence StrengtheningBehavior of HCCS Alloy. Undissolved tertiary carbides sig-nificantly strengthen the HCCS alloy under compression stress at300 �C, in such a way that the cracking that was initiated on grainboundary primary carbides is delayed. The secondary carbidesprobably increase the matrix hardness but have little influence onthe strengthening mechanism itself, regardless of the test tem-perature. Primary carbides promote crack initiation under externalstresses.

4.4 Enhancement of the Softening and the DamagePhenomena Occurring Under Compression at 700 �C

For the CT700-A samples subjected to a high heating rate,the assumption of a distinct prior austenite composition can bemade together with the occurrence of a higher content oftertiary carbides when compared to CT700-B. At the end of thereheating stage, CT700-A may exhibit a microstructure similarto that of the CT300-A samples subjected to the same thermal


cycle. However, this difference within the prior austenitecomposition seems to have little influence on the strengtheningbehavior of the CT700-A samples as their flow curve is similarto that obtained on the CT700-B samples. In fact, all thesamples tested at 700 �C exhibit a similar flow curveregardless of the heating rate used to reach the austenitizationat 950 �C. Therefore, no strengthening effect can be expectedfrom the presence (CT700-A samples) or the lack (CT700-Bsamples) of undissolved tertiary carbides within the matrix.Consequently, the leading mechanism occurring in the 700 �Csamples during the CTs may correspond to the ‘‘softeningdynamic recovery’’ phenomenon, which involves a reductionof dislocation density by mutual annihilation (Ref 47-49). Infact, only a small amount of dislocations may reach the grainboundaries where ‘‘brittle’’ primary carbides are located, priorto their cracking when their strength is exceed by the localstress. Therefore, the carbide micro-cracking observed at700 �C may be similar to the mechanism which occurs at300 �C, with narrow cracks for the higher compressiontemperature and large and branching cracks for the lowercompression temperature.

In conclusion, although high plastic deformation occurswithin the grains associated with significant total elongation,only a limited shearing phenomenon occurs on grain bound-aries, leading to short cracks within primary carbides. The finalresult is a smaller fracture strain at 700 than at 300 �C.

4.5 Link Between the Hardness and the Microstructureof the Studied Samples

It can be seen that both primary and secondary carbides arepresent in all the sample batches. Although these carbides areknown to enhance hardness, no mention will be made of theirpossible influence in what follows as it is assumed that they actin the same manner for all the samples.

4.5.1 AC and HT Samples. The minimum and themaximum of the bulk hardness are, respectively, obtained inthe AC conditions (AC samples) and in the HT conditions (HTsamples). All the other samples exhibit bulk hardness betweenthe previous outer limits (Table 4).

The low hardness achieved in the AC samples is due to thepresence of troostite within the matrix, such a phase beingsofter than bainite and martensite. Conversely, the HT samplesthat exhibit the maximum bulk hardness contain a matrixalmost entirely composed of martensite, the rest being made ofretained austenite. In addition, the martensite present in the HTsamples is probably supersaturated as all the tertiary carbideshave been dissolved during the reheating stage up to 1025 �C,prior to air quenching.

4.5.2 Samples Reheated and Isothermally CompressedDuring the Cooling Stage. As regards the specimensreheated up to 950 �C before being or not being submitted toan isothermal compression stress during the cooling stage, theCT300-B samples exhibit the highest values for both the bulkhardness and the hardness inside the grains.

For the compressed samples, the above results can beexplained as follows. Due to the low reheating rate up toaustenitization, tertiary carbides (h-cementite) are almost dis-solved leading to a supersaturated c1 matrix that has anaustenitic composition with enhanced carbon and alloyingelement contents (CT300-B and CT700-B samples). Converse-ly, h will remain undissolved in the matrix of the samples thathave been submitted to a high reheating rate (CT300-A and

CT700-A samples) yielding a lower carbon-saturated c0 matrix.During the subsequent cooling, the application of compressionstress may promote the bainitic transformation under thedevelopment of preferential oriented sheaves within certaingrains, either at 700 or at 300 �C. The end of the cooling stagedown to room temperature is later achieved following thewithdrawal of the compression stress. During the final coolingstage of the pre-stressed samples, the austenitic grains that werenot transformed into bainite under stress may undergo amartensitic transformation. But either the rate of the martensitictransformation or the hardness of the newly formed martensitewill depend on the chemical composition of the parent austenitephase.

For samples compressed at 300 �C, the supersaturatedc1-austenite present in the CT300-B samples will lead to amartensite with hardness higher than that of the lower carbon-saturated c0-austenite found in the CT300-A samples. Inaddition, the amount of SIBT previously formed at 300 �C willbe higher in the CT300-B than in the CT300-A due to thehigher strain achieved in the former specimen. In fact, the lackof tertiary carbides within the austenitic matrix of the CT300-Bsamples will allow more plastic deformation under stresscompared to the CT300-A samples, which still containundissolved h that strengthens the material. In all, more bainiteis formed in the CT300-B than in the CT300-A samples andalso more martensite exists in the former specimens, thusjustifying why both the bulk hardness and the hardness insidethe grains, with or without bainite, are higher in the CT300-Bthan in the CT300-A samples.

For the samples compressed at 700 �C, a similar microstruc-ture is obtained in the CT700-A and CT700-B samples, leadingto similar hardness values both in the bulk and inside the grains.Very few oriented bainitic sheaves are isothermally formedunder stress at 700 �C compared to 300 �C. After thewithdrawal of the compression stress, a certain amount ofmartensite is formed from the parent deformed austenite, thisamount being higher in grains free of bainite than in grains thatalready contain bainite. Nevertheless, the content of theuntransformed austenite seems to be higher in the CT700samples than in the CT300 samples, probably due to therelative stability of the former that arises from the higher plasticdeformation which occurs at 700 �C compared to 300 �C.Finally, the hardness values achieved in the CT700 samples,which are between those obtained in the CT300-B and theCT300-A samples, are consistent with the microstructurepresent in the former samples.

4.5.3 Sample Reheated and Cooled Down WithoutStress. For the SF300-B sample, the hardness values in thebulk and inside the grains are lower than those obtained in theCT300-B and CT700-B samples that have been subjected tothe same reheating rate. This is because more bainite is formedwithin the SF300-B sample, this bainite exhibiting sheaves thatare haphazardly oriented. Therefore, the bainite present in theSF300-B sample is formed without stress and under continuouscooling conditions, contrary to the SIBT observed in thecompressed samples that have been formed isothermally.

In the SF300-B sample, the large amount of bainite that isformed consumes an equivalent proportion of the prioraustenite phase, which in turn is no more involved in thesubsequent martensitic transformation. Thus, the intrinsichardening potential of the material, which is the maximumwith martensite, is reduced when most of the parent austenitehas been previously transformed into bainite.


4.5.4 Conclusions on the Hardness-MicrostructureRelationship. Both the amount and the chemical compositionof the phases present within the microstructures achieved atroom temperature can justify the hardness values obtainedfollowing the various thermo-mechanical routes taken for thestudied samples.

Higher amounts of martensite within the microstructure leadto higher hardness values both in the bulk and inside the grains.In addition, enhanced carbon content within the martensite,which comes from the supersaturation of the parent austenitephase, will increase the martensite hardness. Conversely,troostite, when present within the microstructure, yields asharp drop in the hardness. This is due to the fact that troostiteis a soft phase, when compared to retained austenite, bainite, ormartensite.

4.6 Rough Damage Model for Internal Crack Initiation

Figure 11 shows a sketch of the internal damage mechanismoccurring in samples submitted to CTs at 300 and 700 �C. Thisrough damage model can be described first depending on thetemperature, and second depending on the type of austeniticmatrix.

When the compression stress is applied at 300 �C on anaustenitic matrix containing undissolved tertiary carbides, highwork hardening rates can be expected, as these ultrafine andcoherent carbides may pin the dislocations while at the sametime impeding their glide toward the grain boundaries.Undissolved tertiary carbides strengthen the matrix in such away that increased compression stresses are necessary to yieldsufficient micro-plastic deformation and new dislocation gen-eration within the grains. In addition, the SIBT which occurswill also account for the strengthening phenomenon. Whendislocations are not pinned by intragranular carbides, then thedislocation glide may be possible in the direction of the grainboundaries where primary carbides are located. The presence ofPFZs close to the grain boundaries may also promotedislocation gliding within these regions. Subsequent pileup ofthe dislocations on the primary carbides leads to a shearingprocess which in turn yields sharp micro-cracks as these speciesare known to be very brittle. The crack initiation within thegrain boundary carbides will then follow the direction of theexternally applied stress.

When the compression stress is applied at 300 �C on asupersaturated austenitic matrix where tertiary carbides havebeen dissolved, the matrix is softer and it can undergosignificant plastic strain under limited externally applied stress.The increase in the compression stress leads to an increase inthe dislocation density within the grains. Due to the weakpresence of fine precipitates having high coherency with thematrix, such as undissolved tertiary carbides, most of the newlygenerated dislocations are free to move toward grain boundarieswhere they will pileup on primary carbides. The crack initiationprocess will then occur in the same manner as for the HCCSmaterial containing undissolved tertiary carbides, with sharpmicro-cracks parallel to the externally applied stress. Similarly,SIBT may also occur under stress. Nevertheless, the appliedstress necessary to initiate cracks within the grain boundarycarbides will be many times lower than that of the samematerial, which contains undissolved tertiary carbides. More-over, the crack path inside the softer material will quicklychange to a more complex and branched one. Such a complexcrack path will contain, in addition to the large cracks parallel

to the direction of the external stress, other micro-cracks whichare not parallel to the external stress.

For CTs performed at 700 �C, the leading phenomenon issoftening dynamic recovery, which is little influenced byintragranular carbides, these precipitates being made either ofthe ultrafine tertiary coherent carbides or of the coarsesecondary carbides. Very little oriented bainitic sheaves areformed at 700 �C, thus confirming the reduced strengtheningphenomenon observed at this temperature. In addition, sig-nificant plastic deformation occurs under stress at 700 �C,compared to the samples compressed at 300 �C.

The quoted damage model (Fig. 11) mainly corresponds tothe internal crack nucleation process on grain boundary primarycarbides (M7C3) under the influence of tertiary carbides (h)present or not within the austenite (c) grains. Conversely, thesecondary intragranular carbides (M23C6) hardly influence boththe strengthening and the softening behaviors of HCCS,probably due to their low coherency relationships with thematrix. Therefore, these carbides are not illustrated on therough damage model.

Note that the start of the martensitic transformation duringcooling is affected both by the carbon content of the parentaustenite and the level of the applied stress, thus explaining thedifferent levels of the martensite start temperatures (MS1, MS2,

and MS3 points) that are drawn in Fig. 11.

5. Conclusions

HCCS alloy after casting and subsequent hardening treat-ment is made of a mixed martensite and retained austenitematrix with both Cr-rich M7C3 and Mo-rich M2C primarycarbides located at the grain boundaries, and widespreadCr-rich M23C6 secondary carbides inside the grains.

When reheating the HCCS alloy up to austenitization, thetertiary carbides precipitate from the decomposition of themartensite before the beginning of the austenite transformation.These carbides may undergo subsequent dissolution in thenewly formed austenite up to 950 �C, especially if low heatingrates are used. High heating rates delay the tertiary carbidedissolution within the austenitic matrix.

Tertiary carbides strongly influence the strengtheningbehavior of the compressed austenite of the HCCS alloy whenthey remain undissolved. Their influence on the strengtheningmechanism is significant at 300 �C where they pin dislocations,but more limited at 700 �C, due to the softening recoveryphenomenon occurring within the material.

Secondary intergranular carbides remain undissolved whenthe HCCS alloy is reheated up to 950 �C. These carbidesincrease the hardness of the matrix inside the grains, butscarcely affect the strengthening behavior of the HCCS alloy.

Primary carbides that are located at grain boundariesrepresent the critical feature in damage mechanism initiation.This is particularly the case for M7C3 carbides that crack due tothe external compression stress.

Compressed austenite, either at 700 or at 300 �C, yieldsbainite with preferentially oriented sheaves, this transformationaccounting also for the strengthening mechanism of the HCCSalloy.

A rough damage model for internal crack initiation isproposed, which takes into account the composition of theparent austenite matrix and the related strengthening and


softening mechanisms occurring under an externally appliedstress.

Acknowledgments

The authors warmly thank Marichal Ketin for supplying thematerial and for their technical support. CAREl of the ULg is alsothanked for providing SEM/EDS facilities.The authors acknowl-edge Conicyt (National Commission for Scientific and Techno-logical Research) Chile, for financial help. InteruniversityAttraction Poles Program-Belgian State-Belgian Science PolicyP7 INTEMATE is thanked for its support. As research Director ofFRS-FNRS, AM Habraken thanks this fund, for financial support.

References

1. A.M. Bayer, B. Becherer, and T. Vasco, High-Speed Tool Steels. ASMHandbook Vol. 16 Machining, ASM International, Materials Park,1989, p 51–59

2. Z. Zhang, D. Delagnes, and G. Bernhart, Microstructure Evolution ofHot-Work Tool Steels During Tempering and Definition of a KineticLaw Based on Hardness Measurements, Mater. Sci. Eng. A, 2004,380(1), p 222–230

3. J. Lecomte-Beckers, M. Sinnaeve, and J. Tchoufang Tchuindjang, NewTrends in Hot Strip Mill Roughing Mills: Characterization of HighChromium Steel and Semi-HSS Grades, Assoc. Iron & Steel Technol.,Vol. 2, 2012, p 1–8

4. K.C. Hwang, S. Lee, and H.C. Lee, Effects of Alloying Elements onMicrostructure and Fracture Properties of Cast High Speed Steel RollsPart I: Microstructural Analysis, Mater. Sci. Eng. A, 1998, 254(1), p282–295

5. M. Boccalini and H. Goldstein, Solidification of High Speed Steels,Inter. Mater. Rev., 2001, 46(2), p 92–115

6. J.W. Park, H.C. Lee, and S. Lee, Composition, Microstructure,Hardness, and Wear Properties of High-Speed Steel Rolls, Met. Mater.Trans. A, 1999, 30(2), p 399–409

7. H. Fu, Y. Qu, J. Xing, X. Zhi, Z. Jiang, M. Li, and Y. Zhang,Investigations on Heat Treatment of a High-Speed Steel Roll, Jrn. Mat.Eng. Perf., 2008, 17(4), p 535–542

8. S.Z. Qamar, Effect of Heat Treatment on Mechanical Properties of H11Tool Steel, Jrn. Ach. Mater. Manuf. Eng., 2009, 35(2), p 115–120

9. I. Neira Torres, G. Gilles, J. Tchoufang Tchuindjang, J. Lecomte-Beckers, M. Sinnaeve, and A.M. Habraken, Study of Residual Stressesin Bimetallic Work Rolls, Adv. Mat. Res., 2014, 996, p 580–585

10. H. Li, Z. Jiang, K.A. Tieu, and W. Sun, Analysis of Premature Failureof Work Rolls in a Cold Strip Plant, Wear, 2007, 263(7), p 1442–1446

11. K.H. Ziehenberger and M. Windhager, Recent Developments in HSMRougher Rolls: Risks and Chances, Iron Steel Technol., 2006, 3(9),p 38–41

12. R. Liang and A.S. Khan, A Critical Review of Experimental Results andConstitutive Models for BCC and FCC Metals Over a Wide Range ofStrain Rates and Temperatures, Int. Jrn. Plast., 1999, 15(9), p 963–980

13. G.Z. Voyiadjis and F.H. Abed, Effect of Dislocation Density Evolutionon the Thermomechanical Response of Metals with Different CrystalStructures at Low and High Strain Rates and Temperatures, Arch.Mech., 2005, 57(4), p 299–343

14. J. Liu, H. Chang, R. Wu, T.Y. Hsu, and X. Ruan, Investigation on HotDeformation Behavior of AISI, T1 High-speed Steel, Mater. Char.,2000, 45(3), p 175–186

15. C.A.C. Imbert and H.J. McQueen, Dynamic Recrystallization of A2and M2 Tool Steels, Mater. Sci. Eng. A, 2001, 313(1), p 104–116

16. I. Tamura, M. Ibaraki, and H. Nozaki, Some Experiments onAusforming of Tool Steels, Trans. JIM., 1996, 7(4), p 248–252

17. D. Jandova, L.W. Meyer, B. Masek, Z. Novy, D. Kesner, and P. Motycka,The Influence of Thermo-mechanical Processing on the Microstructure ofSteel 20MoCrS4, Mater. Sci. Eng. A, 2003, 349(1), p 36–47

18. T. Shigeta, A. Takada, H. Terasaki, and Y.-I. Komizo, The Effects ofAusforming on Variant Selection of Martensite in Cr-Mo Steel, Qart.Jrn. Jpn. Weld. Soc., 2013, 31(4), p 178s–182s

19. P.C.J. Gallagher, The Influence of Alloying, Temperature, and RelatedEffects on the Stacking Fault Energy, Met. Trans., 1970, 1(9), p 2429–2461

20. S.R. Kalidindi, Modeling The Strain Hardening Response of Low SFEFCC Alloys, Int. Jrn. Plast., 1998, 14(12), p 1265–1277

21. U.F. Kocks and H. Mecking, Physics and Phenomenology of StrainHardening: The FCC Case, Prog. Mater. Sci., 2003, 48(3), p 171–273

22. K. Maruyama, K. Sawada, J. Koike, H. Sato, and K. Yagi, Examinationof Deformation Mechanism Maps in 2.25Cr-1Mo Steel by Creep Testsat Strain Rates of 10�11 to 10�6 s�1, Mater. Sci. Eng. A, 1997, 224(1),p 166–172

23. P.J. Ferreira, J.B. VanderSande, M. AmaralFortes, and A. Kyrolainen,Microstructure Development During High-Velocity Deformation, Met.Mater. Trans. A, 2004, 35(10), p 3091–3101

24. J. Tchoufang Tchuindjang, M. Sinnaeve, and J. Lecomte-Beckers,Influence of High Temperature Heat Treatment on in situ Transforma-tion of Mo-rich Eutectic Carbides in HSS and Semi-HSS Grades,Abrasion 2011, Conference Proceedings, Liege, J. Lecomte-Beckers, J.Tchoufang Tchuindjang, Ed., 2011, p 61-75, ISBN 978-2-8052-0124-0

25. J.-Y. Chae, J.-H. Jang, G. Zhang, K.-H. Kim, J.S. Lee, H.K.D.H.Bhadeshia, and D.-W. Suh, Dilatometric Analysis of CementiteDissolution in Hypereutectoid Steels Containing Cr, Scripta Mater.,2011, 65(3), p 245–248

26. I. Neira Torres, G. Gilles, J. Tchoufang Tchuindjang, J. Lecomte-Beckers, M. Sinnaeve, and A.M. Habraken, Prediction of ResidualStresses by FE Simulations on Bimetallic Work Rolls During Cooling,Comp. Meth. Mater. Sci., 2013, 13(2), p 84-91

27. K. Hase, C. Garcia-Mateo, and H.K.D.H. Bhadeshia, Bainite Forma-tion Influenced by Large Stress, Mater. Sci. Technol., 2004, 20(12), p1499–1505

28. H.K.D.H. Bhadeshia, Bainite in Steels: Transformations, Microstruc-tures and Properties, 2nd ed., IOM Communication Ltd, London, 2001

29. J.R. Yang, C.Y. Huang, W.H. Hsieh, and C.S. Chiou, MechanicalStabilization of Austenite Against Bainitic Reaction in Fe-Mn-Si-CBainitic Steel, Mater. Trans. JIM, 1996, 37(4), p 579–584

30. D.S. Liu, G.D. Wang, X.H. Liu, and G.Z. Cui, Mechanical Stabiliza-tion of Deformed Austenite During Continuous Cooling Transforma-tion in a C-Mn-Cr-Ni-Mo Plastic Die Steel, Acta Met. Sin., 2009, 11(2),p 93–99

31. X.J. Jin, N. Min, K.Y. Zheng, and T.Y. Hsu, The Effect of AusteniteDeformation on Bainite Formation in an Alloyed Eutectoid Steel,Mater. Sci. Eng. A, 2006, 438, p 170–172

32. W. Gong, Y. Tomota, Y. Adachi, A.M. Paradowska, J.F. Kelleher, andS.Y. Zhang, Effects of Ausforming Temperature on Bainite Transfor-mation, Microstructure and Variant Selection in Nanobainite Steel,Acta Mater., 2013, 61(11), p 4142–4154

33. H.K.D.H. Bhadeshia, S.A. David, J.M. Vitek, and R.W. Reed, StressInduced Transformation to Bainite in Fe-Cr-Mo-C Pressure VesselSteel, Mater. Sci. Technol., 1991, 7(8), p 686–698

34. H.K.D.H. Bhadeshia, Effect of Stress & Strain on Formation of BainiteSteels, Hot Workability of Steels and Light Alloys-Composites,(Montreal, Canada), H.J. McQueen, E.V. Konpleva, N.D. Ryan, Ed.,Can. Int. Min., 1996, p 543-556

35. S. Kundu, K. Hase, and H.K.D.H. Bhadeshia, Crystallographic Textureof Stress-affected Bainite, Proc. R. Soc. A, 2007, 463(2085), p 2309–2328

36. A.A. Shirzadi, H. Abreu, L. Pocock, D. Klobcar, P.J. Withers, andH.K.D.H. Bhadeshia, Bainite Orientation in Plastically DeformedAustenite, Int. Jrn. Mater. Res., 2009, 100(1), p 40–45

37. M. Durand-Charre, Microstructure of Steels and Cast Irons, Chap. 11,Springer, New York, 2004, p 219–222, ISBN 3-540-20963-8

38. D.V. Shtansky, K. Nakai, and Y. Ohmori, Decomposition of Martensiteby Discontinuous-like Precipitation Reaction in an Fe-17Cr-0.5CAlloy, Acta Mater., 2000, 48(4), p 969–983

39. G. Miyamoto, H. Usuki, Z.-D. Li, and T. Furuhara, Effects of Mn, Siand Cr Addition on Reverse Transformation at 1073 K fromSpheroidized Cementite Structure in Fe-0.6 Mass% C Alloy, ActaMater., 2010, 58(13), p 4492–4502

40. K.D. Clarke, C.J. Van Tyne, C.J. Vigil, and R.E. Hackenberg, InductionHardening 5150 Steel: Effects of Initial Microstructure and HeatingRate, Jrn. Mater. Eng. Perf., 2011, 20(2), p 161–168

41. S.-J. Lee and K.D. Clarke, A Conversional Model for AusteniteFormation in Hypereutectoid Steels, Met. Mater. Trans. A, 2010,41(12), p 3027–3031


42. M. Dunckelmeyer, C. Krempaszky, E. Werner, G. Hein, and K.Schorkhuber, On the Causes of Banding Failure, Proc. of METECInSteelCon 2011, STEELSIM, 2011, p 1–9

43. P.W. Shelton and A.S. Wronski, Cracking in M2 High Speed Steel,Mater. Sci. Technol., 1983, 17(11), p 533–540

44. M. Kroon and J. Faleskog, Micromechanics of Cleavage FractureInitiation in Ferritic Steels by Carbide Cracking, Jrn. Mech. Phys. Sol.,2005, 53(1), p 171–196

45. C.R.S. da Silva and M. Boccalini, Thermal Cracking of Multicompo-nent White Cast Iron, Mater. Sci. Technol., 2005, 21(5), p 565–573

46. B.Z. Margolin, V.A. Shvetsova, and A. Ya, Varovin, PreliminaryCompression of a Material as a Factor in Changing the Brittle FractureMechanism for BCC Metals, Str. Mater., 1996, 28(4), p 251–261

47. V. Vodopivec, Dynamic Recovery of Austenite in Low Carbon Steelsand its Relationship to the Precipitation of AlN, Jrn. Mater. Sci., 1975,10(6), p 1082–1084

48. E. Nes, Recovery Revisited, Acta Mater., 1995, 43(6), p 2189–220749. L. Kubin, T. Hoc, and B. Devincre, Dynamic Recovery and its

Orientation Dependence in Face-centered Cubic Crystals, Acta Mater.,2009, 57(8), p 2567–2575


Phase Transformations and Crack Initiation in a High ... · Phase Transformations and Crack...

Documents

Transcript of Phase Transformations and Crack Initiation in a High ... · Phase Transformations and Crack...